Relevancy Configurations: Default Search Operator, Special Character Search, Neural Search, Hybrid Search
The Relevancy tab offers four settings:
-
Search Operator, where you can define the default relationship between two and more keywords.
-
Special Characters Search, where you can control how nonalphanumeric characters are processed during search.
-
Neural Search is a toggle. Turn it on to move from keyword search to neural search and turn it off move from neural search back to keyword search.
-
Hybrid Search is a toggle. Turn it on to use both keyword search and neural search simultaneously. Turn it off when you want to use either keyword search or neural search.
You can toggle on Neural Search or Hybrid as per the requirement. For more information on different search methods please see Difference between keyword , Neural and Hybrid search
Search Operator
This configuration offers two settings: OR and AND. When OR is selected, a query "search clients" is interpreted as "search OR clients." Documents that are returned contain either search or client or both. When AND is selected, a query "search clients" is interpreted as "search AND clients." Only the documents containing both "search" and "client" are returned.
-
In Search Clients, open a search client for editing.
-
Jump to the Relevancy tab.
-
From the first dropdown, Search Operator, select OR or AND.
-
Save your settings.
Special Characters Search
When used with an #
in queries such as # search + client
, +
acts an advanced search operator. The complete list of such operators can be found on Use Advanced Search. The exclusive use of + as an operator creates a problem. What if your users want to find documents on, let's say, C++
? To cater to such users, you can turn on Special Characters Search.
-
In Search Clients, open a search client for editing.
-
Click the Relevancy tab.
-
Toggle on Special Character Search.
-
Save your settings.
When Special Character Search is on, end users can look up the following characters searchable.
№ | Searchable Character |
1 | Ampersand (&) |
2 | At Sign (@) |
3 | Greater Than (>) |
4 | Less Than (<) |
5 | Dollar ($) |
6 | Percentage (%) |
7 | Caret (^) |
8 | Equals (=) |
9 | Colon (:) |
10 | Full stop(.) |
11 | Forward slash (/) |
12 | Hyphen (-) |
13 | Plus (+) |
14 | Asterisk (*) |
Any of these characters can be searched outside advanced search queries; 5 > 2
works but # 5 > 2
and # 5 > 2 OR 3
don't. There is an exception to the rule. Special characters can work inside advanced search when they are enclosed in quotes. So, # "5 > 2"
works.
Neural Search
Neural Search is an information retrieval method with a focus on meaning. Unlike Keyword Search, which returns results based on their similarity to the search query, Neural Search returns documents based on their semantic closeness to the search query. A query, such as, "best movies of 2024" can return results which contain none of the words from the query but are relevant to the user intention. To learn more, visit About Neural, Hybrid and Keyword Search.
You can activate or deactivate Neural Search with a toggle. However, to learn what happens when you turn in on and how content sources and search client configuration changes impact Neural Search, read Full Neural Search (with Hybrid)
Hybrid Search
Hybrid Search can prioritize Keyword Search or Neural Search based on the query. When Hybrid Search is active, then queries identified as exact match get results from the keyword search algorithms and broad queries rely on neural search algorithms. You can learn more about all three kinds of search mechanisms on About Neural, Hybrid and Keyword Search
To use Hybrid Search, toggle it on. When Hybrid Search is on, then vectorization takes place once a week. But if you have turned on Hybrid Search for the first time, then vectorization will start immediately. Results will be powered by Hybrid Search after the vectorization is successful and complete. However, toggling off Hybrid Search has an immediate impact on search. More information is on Hybrid Search.